LaTeX Notebook Functionality

This package contains tools for running a LaTeX notebook with embedded python code. One can then, for instance, keep an electronic lab notebook where the plots and data are generated and processed in place, immediately from the raw data. (The author of the project has kept an electronic notebook for several years in this way with great success.)

Please note that this is a very different “notebook” than the one supplied by jupyter notebooks (which pyspecdata also supports). In this section, we aim to provide a laboratory notebook with publication-ready figures and a complete path from raw data to those figures.

  • With minimal effort, it is possible to design a notebook that works in a similar fashion with HTML, Markdown, etc. (The main author just has no interest in doing this, since the PDF output looks very nice.)
  • It’s highly recommended to keep a notebook as a series of different files that have no preamble or ending (i.e. only the part that goes inside the document environment), which can then be collected into a gigantic master document with the \input{...} command, or compiled individually (e.g. while you are actually in the lab working on a particular section).
  • It’s also highly recommended to store notes organized by project, which can then be cross-referenced in a separate chronological document (or vice versa) with the \\ref{...} command.

Setting up the notebook

Requirements

  1. latex packages

    You need to put the texmf tree in a location where it can be found by your latex installation.

    • Under Windows, add the texmf tree to “miktex settings” under the “roots” tab.
    • (To do) We should use this guide or or this package to package the code and include it here.

    Once you’ve done this, the shell command kpsewhich mypython.sty should return a result (if you have miktex installed on windows, this should work from either the git bash prompt or the dos or powershell prompt).

  2. The pyspecdata package.

    Proves the commands pdflatex_notebook_wrapper and update_notebook_pythonscripts, described below under “Running the notebook.”

    Also provies the command pdflatex_notebook_view_wrapper, which is used to determine the output PDF and call an appropriate viewer.

  3. A standard latex compilation system:

    You can use latexmk (shipped with miktex) with Sumatrapdf (Sumatrapdf allows you to edit the PDF while it’s open in Sumatrapdf, while Adobe Acrobat does not). Here is a ~/.latexmkrc file that works on windows:

    $pdflatex=q/pdflatex_notebook_wrapper %O -synctex=1 --xelatex %S/;
$pdf_previewer=q/pdflatex_notebook_view_wrapper/;#calls the wrapviewer function

    It should also be possible to use TeXworks by adding pdflatex_notebook_wrapper to preferences → typesetting → processing tools.

    Todo

    Alec, can you check this out and update documentation??

  4. It’s assumed that your latex files are stored in a “notebook directory.” In some cases, during the first run, an explanatory error will appear – just follow the instructions.

Running the notebook

If you create figures with the figlist_var class, you should simply be able to write a latex file with embedded python environments (\\begin{python} … \\end{python}) replace the pdflatex command with pdflatex_notebook_wrapper when compiling your latex notebook, to drop the code and plots in place. For clarity, the code output is a slightly different color (a dark brown) than the standard text.

A synctex “jump to source” on the resulting portion of the PDF will send you to the tex output, which is stored in scripts/*.tex, where * is a sequential number corresponding to the script, and the python source used to generate it is stored in scripts/*.py.

Each snippet of unique code is run only once, ever making the notebook fast and efficient. For now, an important drawback to this is that if you change modules or libraries called by the script, the resulting PDF output will not change. To get around this, a command is provided that forces scripts to be re-run. You use it like this: update_notebook_pythonscripts flush 10 21 – which will flush script numbers 10 to 21. Manually deleting the .py files inside the scripts directory will not have the same effect.

Todo

To limit downtime for the PDF, pdflatex_notebook_wrapper currently copies the final pdf to a truncated filename (assuming that the filename consists of words separated by underscores, it drops the last word).

It would be much better to copy the source PDF into a subdirectory, build it there, and then copy the pdf back into the main directory. This would entail changing the paths of the various files

\RequirePackage[log]{snapshot} might be helpful to log files here.

probably we will just want to add commands to renewcommand for input as well as the graphicx root.

How it works

  • Note that the code works a bit differently than in previous versions – while it previously required LaTeX to be run with shell-escape enabled, pdflatex_notebook_wrapper doesn’t require this.
  • pdflatex_notebook_wrapper just calls pdflatex followed by update_notebook_pythonscripts
  • To compile, use pdflatex_notebook_wrapper --xelatex

The LaTeX end

The file mypython.sty looks for the python environment, it pulls the relevant code, outputs it to scripts/*.py and then writes a command to the .aux file that tells LaTeX where to find the scripts/*.tex output. The scripts/*.tex output is only updated once update_notebook_pythonscripts (without arguments) is run.

The python end

update_notebook_pythonscripts runs through the various scripts/*.py files, checks and checks whether or not they have been previously cached (under the same or a different script number). If the python code is in the cache, it just pulls the cached output. If not, it runs the file, and stores the result in the cache.